BLP File

Introduction

BLP files are used as texture storage. The textures can be stored with a 256 color palette or full 24bit RGB colors. The format supports 1, 4 and 8-bit alpha transparency and DXT compression. The file format is NOT chunked. Wikipedia has a nice overview over the format: http://en.wikipedia.org/wiki/.BLP

Header

From http://www.pxr.dk/wowdev/wiki/index.php?title=BLP

Offset Type        Description
------------------------------------------------------------------------------------------
0x00   char[4]     always 'BLP2'
0x04   uint32      Type, always 1
0x08   uint8       Compression: 1 for uncompressed, 2 for DXTC
0x09   uint8       Alpha channel bit depth: 0, 1 or 8
0x0A   uint8       Alpha encoding 0x0B uint8 Has MipMaps? 
0x0C   uint32      X resolution (power of 2)
0x10   uint32      Y resolution (power of 2)
0x14   uint32[16]  offsets for every mipmap level (or 0 when there is no more mipmap level)
0x54   uint32[16]  sizes for every mipmap level (or 0 when there is no more mipmap level)
0x94   uint32[256] palette of 256 BGRA Values

If HasMipMaps is 0, there is only 1 mipmap level. The palette is always present, even if it is not used. In that case all values are 0.

Encoding Schemes

Type 1 Compression 1 AlphaDepth 0 (uncompressed paletted image with no alpha)

Each byte of the image data is an index into Palette which contains the actual RGB value for the pixel. Although the palette entries are 32-bits, the alpha value of each Palette entry may contain garbage and should be discarded.

Type 1 Compression 1 AlphaDepth 1 (uncompressed paletted image with 1-bit alpha)

This is the same as Type 1 Encoding 1 AlphaDepth 0 except that immediately following the index array is a second image array containing 1-bit alpha values for each pixel. The first byte of the array is for pixels 0 through 7, the second byte for pixels 8 through 15 and so on. Bit 0 of each byte corresponds to the first pixel (leftmost) in the group, bit 7 to the rightmost. A set bit indicates the pixel is opaque while a zero bit indicates a transparent pixel.

Type 1 Compression 1 AlphaDepth 8 (uncompressed paletted image with 8-bit alpha)

This is the same as Type 1 Encoding 1 AlphaDepth 0 except that immediately following the index array is a second image array containing the actual 8-bit alpha values for each pixel. This second array starts at BLP2Header.Offset[0] + BLP2Header.Width * BLP2Header.Height.

Type 1 Compression 2 AlphaDepth 0 (DXT1 no alpha)

The image data are formatted using DXT1 compression with no alpha channel.

Type 1 Compression 2 AlphaDepth 1 (DXT1 one bit alpha)

The image data are formatted using DXT1 compression with a one-bit alpha channel.

Type 1 Compression 2 AlphaDepth 4 AlphaEncoding 1 (DXT3 four bits alpha)

The image data are formatted using DXT3 compression.

Type 1 Compression 2 AlphaDepth 8 AlphaEncoding 1 (DXT3 eight bits alpha)

The image data are formatted using DXT3 compression.

Type 1 Compression 2 AlphaDepth 8 AlphaEncoding 7 (DXT5)

The image data are formatted using DXT5 compression.

DXT Compression

BLP only uses DXT 1,3 and 5. From: http://en.wikipedia.org/wiki/DXTn

DXT1

DXT1 (also known as Block Compression 1 or BC1) is the smallest variation of S3TC, storing 16 input pixels in 64 bits of output, consisting of two 16-bit RGB 5:6:5 colour values and a 4x4 two bit lookup table.

If the first colour value (c₀) is numerically greater than the second colour value (c₁), then two other colours are calculated, such that

c₂ = 2/3 c₀ + 1/3 c₁

and

c₃ = 1/3 c₀ + 2/3 c₁

Otherwise, if c₀ <= c₁, then

c₂ = 1/2 c₀ + 1/2 c₁

and c₃ is transparent black corresponding to a pre-multiplied alpha format.

The lookup table is then consulted to determine the colour value for each pixel, with a value of 0 corresponding to c₀ and a value of 3 corresponding to c₃ . DXT1 does not store alpha data enabling higher compression ratios.

DXT3

DXT3 (also known as Block Compression 2 or BC2) converts 16 input pixels (corresponding to a 4x4 pixel block) into 128 bits of output, consisting of 64 bits of alpha channel data (4 bits for each pixel) followed by 64 bits of colour data, encoded the same way as DXT1 (with the exception that the 4 colour version of the DXT1 algorithm is always used instead of deciding which version to use based on the relative values of c₀ and c₁ ).

In DXT3, the colour data is interpreted as not having been pre-multiplied by alpha. Typically DXT2/3 are well suited to images with sharp alpha transitions, between translucent and opaque areas.

DXT5

DXT5 (also known as Block Compression 3 or BC3) converts 16 input pixels into 128 bits of output, consisting of 64 bits of alpha channel data (two 8 bit alpha values and a 4x4 3 bit lookup table) followed by 64 bits of colour data (encoded the same way as DXT2 and DXT3).

If α₀ > α₁, then six other alpha values are calculated, such that

α₂ = (6α₀ + 1α₁) / 7,

α₃ = (5α₀ + 2α₁) / 7,

α₄ = (4α₀ + 3α₁) / 7,

α₅ = (3α₀ + 4α₁) / 7,

α₆ = (2α₀ + 5α₁) / 7,

α₇ = (1α₀ + 6α₁) / 7

Otherwise, if α₀ <= α₁, four other alpha values are calculated such that

α₂ = (4α₀ + 1α₁) / 5,

α₃ = (3α₀ + 2α₁) / 5,

α₄ = (2α₀ + 3α₁) / 5,

α₅ = (1α₀ + 4α₁) / 5,

α₆ = 0,

α₇ = 255

The lookup table is then consulted to determine the alpha value for each pixel, with a value of 0 corresponding to α₀ and a value of 7 corresponding to α₇. DXT5's colour data is not pre-multiplied by alpha. Because DXT4/5 use an interpolated alpha scheme, they generally produce superior results for alpha (transparency) gradients than DXT2/3.